From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757408AbcCXLIX (ORCPT ); Thu, 24 Mar 2016 07:08:23 -0400 Received: from mail-wm0-f48.google.com ([74.125.82.48]:37540 "EHLO mail-wm0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755585AbcCXLIO (ORCPT ); Thu, 24 Mar 2016 07:08:14 -0400 Subject: Re: [PATCH] KVM: x86: optimize steal time calculation To: Liang Chen References: <1458127996-21799-1-git-send-email-liangchen.linux@gmail.com> Cc: kvm@vger.kernel.org, linux-kernel@vger.kernel.org, Gavin Guo From: Paolo Bonzini Message-ID: <56F3CA9B.9070403@redhat.com> Date: Thu, 24 Mar 2016 12:08:11 +0100 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.6.0 MIME-Version: 1.0 In-Reply-To: <1458127996-21799-1-git-send-email-liangchen.linux@gmail.com> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 16/03/2016 12:33, Liang Chen wrote: > Since accumulate_steal_time is now only called in record_steal_time, it > doesn't quite make sense to put the delta calculation in a separate > function. The function could be called thousands of times before guest > enables the steal time MSR (though the compiler may optimize out this > function call). And after it's enabled, the MSR enable bit is tested twice > every time. Removing the accumulate_steal_time function also avoids the > necessity of having the accum_steal field. > > Signed-off-by: Liang Chen > Signed-off-by: Gavin Guo > --- > arch/x86/include/asm/kvm_host.h | 1 - > arch/x86/kvm/x86.c | 19 +++---------------- > 2 files changed, 3 insertions(+), 17 deletions(-) > > diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h > index 44adbb8..3f460a7 100644 > --- a/arch/x86/include/asm/kvm_host.h > +++ b/arch/x86/include/asm/kvm_host.h > @@ -547,7 +547,6 @@ struct kvm_vcpu_arch { > struct { > u64 msr_val; > u64 last_steal; > - u64 accum_steal; > struct gfn_to_hva_cache stime; > struct kvm_steal_time steal; > } st; > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index eaf6ee8..82bfefc 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -2002,22 +2002,8 @@ static void kvmclock_reset(struct kvm_vcpu *vcpu) > vcpu->arch.pv_time_enabled = false; > } > > -static void accumulate_steal_time(struct kvm_vcpu *vcpu) > -{ > - u64 delta; > - > - if (!(vcpu->arch.st.msr_val & KVM_MSR_ENABLED)) > - return; > - > - delta = current->sched_info.run_delay - vcpu->arch.st.last_steal; > - vcpu->arch.st.last_steal = current->sched_info.run_delay; > - vcpu->arch.st.accum_steal = delta; > -} > - > static void record_steal_time(struct kvm_vcpu *vcpu) > { > - accumulate_steal_time(vcpu); > - > if (!(vcpu->arch.st.msr_val & KVM_MSR_ENABLED)) > return; > > @@ -2025,9 +2011,10 @@ static void record_steal_time(struct kvm_vcpu *vcpu) > &vcpu->arch.st.steal, sizeof(struct kvm_steal_time)))) > return; > > - vcpu->arch.st.steal.steal += vcpu->arch.st.accum_steal; > + vcpu->arch.st.steal.steal += current->sched_info.run_delay - > + vcpu->arch.st.last_steal; > + vcpu->arch.st.last_steal = current->sched_info.run_delay; > vcpu->arch.st.steal.version += 2; > - vcpu->arch.st.accum_steal = 0; > > kvm_write_guest_cached(vcpu->kvm, &vcpu->arch.st.stime, > &vcpu->arch.st.steal, sizeof(struct kvm_steal_time)); > Nice cleanup; queued for 4.7, thanks. Paolo